Stepping up competition in the e-book subscription category, Scribd is launching a new browsing feature, going live today, that claims to combine the best of human and algorithmic book recommendation.

The new browsing feature is intended to distinguish Scribd's e-book subscription service—through which consumers can read an unlimited number of books for $8.99 a month—from competitors like Oyster and the recently launched Kindle Unlimited.

Scribd co-founder and CTO Jared Friedman said the company is putting an emphasis on book discovery because of the unique characteristics of digital subscriptions. “Making books free to read creates new consumer behavior. We see an opportunity to take book discovery to a new level on Scribd. When people go to a physical bookstore they usually know what they want; on Scribd people discover books they didn’t know about before, and read them.”

Over the last year Scribd has developed a new browsing infrastructure that Friedman said was based on “our favorite physical bookstores.” The new browsing structure includes three major features: it combines human selectivity and computer algorithms; it offers deep curation for readers of every taste; and it allows for personalization, making the browsing experience better the more a consumer uses it.

Friedman said the new infrastructure was produced by a team of 50 developers (including data scientists, librarians, publishing professionals and more) and offers an increased number of subject categories to drive recommendations and curation. The new service also features “designated spaces” to promote specific titles and offer publishers “clearly labeled” marketing opportunities.

Now, clicking on the browse feature gives you a page offering Scribd Selects (curatorial recommendations form the Scribd staff) and Top Books (whatever books are trending on the site); below these top level categories are a page of genre categories to choose from, including Biography and Memoir, Humor, Young Adult, Romance and many more.

Within each of these genre categories there are specific curated staff recommendations, in turn followed by five algorithmic recommendations. Below these two levels there is an ever-increasing number of large topical subject collections—each offering 25 or more titles—that include both human and algorithmic-determined title suggestions. These topical collections can be something like, “Lives of the Comedians,” in the humor section or “Mankind’s Feats and Follies,” in history.

Friedman said Scribd’s browsing infrastructure is based on creating its own “proprietary category structure,” thousands of newly created categories, far more than the 1800 book categories offered by BISAC—“readers are varied and BISAC genre categories don’t always match a readers taste,” Friedman said. He also emphasized that the categories (and the collections built from them) constantly change based on tracking user reading behavior and the ongoing addition of new, tagged titles to the Scribd e-book subscription inventory.

Scribd cofounder and CEO Trip Adler added that the company's "model does wonders for book discovery, because it eliminates so much of the friction that usually exists to begin reading a good book.”

“You can use a lot of modalities to find books,” Friedman told PW about the new browsing service. “Readers have their own ways to discover books. We believe we can drive the distribution of backlist and lesser known titles.”